Troubleshooting

The short list of things that go wrong, and the fix for each.

An employee seems stuck

  • Check the Runs page first. If there's a Needs action row, the employee is waiting on you, not stuck.
  • Check the status bar above the chat composer. If it says Paused, resume it.
  • A long silent stretch usually means a long step, like a big page read or a slow site. The run's timeline shows the current step.
  • If it's genuinely wedged, hit stop and re ask with tighter instructions. The run history keeps everything for the post mortem.

An app won't connect

  • Connections open in a new tab; a popup blocker can eat it. Allow popups for useverse.ai and try again.
  • Finish the whole approval flow on the app's page, including picking the account. Closing early leaves the connection half made.
  • If the card still shows Connect after you approved, refresh the page; if it persists, disconnect and reconnect once.
  • Corporate Google or Microsoft accounts sometimes need an admin to allow new apps. The error page from the provider will say so.

Voice call problems

  • The employee can't hear you: check the browser's mic permission for useverse.ai, and that the right input device is selected in your system settings.
  • You can't hear it: check Mute speaker isn't on, and your output device.
  • The call won't connect: use the Retry connection button, and if it keeps failing, refresh the page and start the call again.

A scheduled run didn't fire

  • For a workflow: confirm the Active toggle is on. Drafts never run.
  • For an employee trigger: confirm the employee isn't paused.
  • App event triggers that poll can lag a minute or two behind the event. That's normal.
  • Check the run history for a Skipped or Failed entry; the detail view says why.

A workflow won't activate

Activation validates every step. The message lists exactly which nodes are unconfigured; open each one, fill the required fields, and look for the green check. On the free plan, activation itself is the locked feature, so this is also the moment you'll be asked to upgrade.

Repeated failures on one task

  • Open the failed run and read the timeline; the failing step is marked.
  • If it's an app step, disconnect and reconnect that app. Expired access is the usual suspect.
  • Check the employee's memory for something wrong or stale, and fix it there. See Memory.
  • Still failing? Bring the run to the Discord or support; the run's date and the employee's name are enough for us to find it.

Desktop app: This Mac isn't showing as connected

  • Make sure you're signed in inside the desktop app itself, then restart it. Registration happens on sign in.
  • The per employee toggle matters too: Let this employee work on my computer lives in that employee's Settings tab, and it's off by default.
  • For screen control on macOS, grant Accessibility and Screen Recording in System Settings, then restart the app.

Emails aren't going out

  • Check the employee's limits: Confirm before external emails turns sends into approval requests, which wait for you.
  • Check Allowed email domains; a recipient outside the list is refused by design.
  • The hourly outbound cap queues extra messages rather than dropping them.

Tip

When reporting a bug, include what you expected, what happened, roughly when, and the employee or workflow involved. A screenshot of the run's technical details is gold.

© 2026 Verse. All rights reserved.

Privacy Policy and Terms of Service